Pushing the limits of raw waveform speaker recognition

Abstract

In recent years, speaker recognition systems based on raw waveform inputs have received increasing attention. However, the performance of such systems are typically inferior to the state-of-the-art handcrafted feature-based counterparts, which demonstrate equal error rates under 1% on the popular VoxCeleb1 test set. This paper proposes a novel speaker recognition model based on raw waveform inputs. The model incorporates recent advances in machine learning and speaker verification, including the Res2Net backbone module and multi-layer feature aggregation. Our best model achieves an equal error rate of 0.89%, which is competitive with the state-of-the-art models based on handcrafted features, and outperforms the best model based on raw waveform inputs by a large margin. We also explore the application of the proposed model in the context of self-supervised learning framework. Our self-supervised model outperforms single phase-based existing works in this line of research. Finally, we show that self-supervised pre-training is effective for the semi-supervised scenario where we only have a small set of labelled training data, along with a larger set of unlabelled examples.

Disentangled Speaker Representation Learning via Mutual Information Minimization

“…In the embedding extraction step, audio with variable duration is converted into a single fixed-dimensional vector representation called speaker embedding, which is assumed to contain speakerrelevant information. With a sophisticated speaker embedding, even a simple scoring method such as cosine similarity or euclidean distance has shown high speaker verification performance [2]- [4]. Therefore, most studies have been focused on how to extract a fine speaker embedding from input speech.…”